Postoperative hypothermia increases the incidence of ischaemic cardiac events in patients at risk, but the underlying mechanism is unclear. One possibility is increased cardiac work related to the sympathoneural or adrenomedullary hormonal responses. In awake human volunteers, the present study assessed the effects of mild core hypothermia on these responses, and on the associated changes in indices of cardiac work. A total of 11 healthy men were studied on two separate days. On one day, core temperature (T(c)) was decreased by the intravenous infusion of cold normal saline (4 degrees C; 60 ml/kg over 30 min) through a central venous catheter. On the other day (normothermic control), warm normal saline (37 degrees C; 60 ml/kg over 30 min) was given intravenously. Transthoracic echocardiograms, the sympathoneural response (noradrenaline) and the adrenomedullary response (adrenaline) were evaluated before, during and after the intravenous infusions. Echocardiography was used to measure left ventricular function and cardiac output. Compared with the normothermic control treatment, core cooling of 0.7 degrees C was associated with increased plasma noradrenaline (220% increase; P=0.001), whereas adrenaline, cardiac output, heart rate and the rate-pressure product were unchanged. After core cooling of 1.0 degrees C, increases in noradrenaline (by 230%; P=0.001), adrenaline (by 68%; P=0.05), cardiac output (by 23%; P=0.04), heart rate (by 16%; P=0.04) and rate-pressure product (by 25%; P=0.007) were all significant compared with the normothermic control treatment. In conclusion, there is a T(c) threshold, below which an adrenomedullary (adrenaline) response is triggered in addition to the sympathoneural (noradrenaline) response. This T(c) threshold (approximately 1 degrees C below the normothermic baseline) is also associated with an increase in haemodynamic indices of cardiac work. Mild core hypothermia therefore constitutes a catecholamine-mediated cardiovascular "stress test".